Character Generation

Hi! My first post, and naturally it’s about numbers.

I did some random sampling (~500,000 each), and it looks like the Advanced Character generation methods have these average results:

Adventurer: 11.90
Conqueror: 12.51
King: 12.87

These all assume optimal behavior, which of course, not all players will follow. And the averages do not account for -2/+1 optimization, although I suspect that would drop all of them by about -0.5 on average (so 11.40, 12.01, and 12.47 respectively).

I tend to be a soft-hearted GM, not old-school at all, so I would probably go with the King method at 1st level, but that is so tantalizingly similar to the results of 4d6, drop the lowest, that I think I would rather just do the 4d6 method (still in order) and let the player increase one score by +2 (max 18) or two scores by +1 (max 18) to get the averages roughly correct.

Thomas - thanks very much for doing the random sampling. I don't have the ability to run such large random samples readily. Can I hit you up if I need some data analysis on other projects?!

Sure :-). Volunteer work is subject to spare time constraints, of course, but otherwise I’m happy to help.


I had my group generate their ability scores using the King method, and most of them did -2/+1 where necessary to ensure the +10% XP. Adding up their scores, I get an average of … 12.43. Right on target. Unfortunately, I don’t have their original rolls anymore to compare that part.

I think you nailed it, at least for my group.

I think I prefer rolling five times and picking over 4d6, though, even if the end results are so similar. There’s something about having a handful of potential to choose from.

Thanks to this analysis, though, I may allow the players to try the 4d6 method, though, if they like that better. As long as it’s statistically indistinguishable, let fun win out.

The 3d6 in order thing is probably my only hard sticking point with the rules. I understand the idea behind it: It’s game about normal people going out and doing great things and becoming powerful through their acts, not about superheros going out and beating everything up.

Still, though, I think I look at chargen from a different perspective. There are above-average people in the world. Most of them are above average in one way, and below average in another way. When choosing whose role I want to assume, I would tend to pick one of these exceptional people instead of a pig farmer. Why shouldn’t I be able to?

A good example comes from a real world article I read today:

Here’s a person who clearly has a strength of 18, and a charisma of probably 7 or 8. From her employment situation, I would guess that her intelligence is 10 at best. (Else she’d hopefully be able to find a way to leverage her abilities to find income.) In other words, a person way above average in some ways, totally average in some ways, and below average in other ways. But an admirable person, and one who a person playing the roleplaying game which is our universe might choose to portray. So they sit down ready to play the role of Sarah Robles. But none of the player’s five characters happen to describe such a person. Sorry. You can’t play that character.


This is why I like more flexible systems. Personally, I’ve always held to rolling 4d6 two times, dropping the lowest die of each roll, and 3d6 four times, and assign the resulting 6 rolls as desired. Now you will get a slightly more powerful character, but it allows you more control over who it is you end up portraying. In cases where characters need to be generated away from the Judge’s presence, point buy systems allow the same customization while allowing the Judge to influence the average power of a character by adjusting the number of points available. Both ways work great.

If you want a truly average person, but still want more control over who you get to play, then use 3d6 only - but assign the six scores in whatever order you desire.

I’d love to hear input from the creators on why this would be detrimental to the game.

Hi Mr. Grogg,

It's not detrimental per se. All of the mechanics you've described are fine. I've personally run ACKS with everything from straight 3d6 to the crazy Unearthed Arcana system of 9d6, 8d6, etc. Chargen is very much a matter of taste.

I will say that I think point-buy is not a good idea for games that aim to have any sense of simulation. Random ability scores result in a world wherein not every character conforms to stereotype, in a way that point buy systems never do.

Let me give an example of what I mean. Assume you are creating a fighter, and you get to assign 18, 16, 14, 13, 10, 10 as your attributes. Here is what 90% of players will create:

S 18, C 16, D 14, I 10, W 13 Ch 10; or S 18, C 14, D 16, I 10, W 13, Ch 10; or S 14, C 16, D 18, I 10, W 13, Ch 10 (for the archer lovers). They might move the 13 to Ch or Int. But that's basically how the stats are going to be allocated.

This trend became painfully obvious in 3.5, when players calculated the mathematically best placement of ability scores using the normal array, elite array, and each level of point buy. Optimizers would routinely chide players for sub-optimal allocation. Etc. 

With purely random ability scores, though, you might get a character who looks like this:

S 16, C 14, D 13, I 10, W 10, Ch 18; or

S 13, C 18, D 10, I 16, W 13, Ch 10; or

S 18, C 10, D 10, I 13, W 10, C 16; and so on

It makes the game world richer and more realistic if sometimes you have a fighter who's clumsy but charismatic; or not as strong, but very cunning; or iron-willed and charismatic, but with a weak constitution; and so on. Such persons populate our history books, yet - in the absence of random rolls - they will never populate our game worlds.

I agree with you that our PCs/adventurers should represent the better-than-average human being. ACKS' recommended system is to roll 3d6 in order, generating 5 characters, and then pick the best 1. That gives you a character who is in the top 20% of the human race, but it emerges organically, rather than by fiddling with the dice.

If you want to be more heroic, roll up 10 character, and pick the best 1. For "chosen one" campaigns, roll up 100 characters and pick the best 1. (With an automated character generator, it's pretty easy to do that sort of thing).